Geometric Neural Phrase Pooling: Modeling the Spatial Co-occurrence of Neurons

TL;DR

This paper introduces Geometric Neural Phrase Pooling (GNPP), a novel method for modeling spatial co-occurrence of neuron responses in CNNs, leading to improved image classification accuracy.

Contribution

It proposes GNPP, a new pooling layer that encodes neural phrases based on spatial relationships, enhancing CNN performance with minimal computational cost.

Findings

GNPP improves classification accuracy across multiple datasets.

GNPP effectively models spatial co-occurrence of neuron responses.

The method introduces minimal additional computational overhead.

Abstract

Deep Convolutional Neural Networks (CNNs) are playing important roles in state-of-the-art visual recognition. This paper focuses on modeling the spatial co-occurrence of neuron responses, which is less studied in the previous work. For this, we consider the neurons in the hidden layer as neural words, and construct a set of geometric neural phrases on top of them. The idea that grouping neural words into neural phrases is borrowed from the Bag-of-Visual-Words (BoVW) model. Next, the Geometric Neural Phrase Pooling (GNPP) algorithm is proposed to efficiently encode these neural phrases. GNPP acts as a new type of hidden layer, which punishes the isolated neuron responses after convolution, and can be inserted into a CNN model with little extra computational overhead. Experimental results show that GNPP produces significant and consistent accuracy gain in image classification.